Volume 18, Issue 4 - Special Issue of the 1st International Applied Bioinformatics Conference (iABC'21); Guest Editors: Jens Allmer, Mourad Elloumi, Matteo Comin, and Ralf Hofestädt

Sequencing technologies has provided the basis of most modern genome sequencing studies due to its high base-level accuracy and relatively low cost. One of the most demanding step is mapping reads to the human reference genome. The reliance on a single reference human genome could introduce substantial biases in downstream analyses. Pangenomic graph reference representations offer an attractive approach for storing genetic variations. Moreover, it is possible to include known variants in the reference in order to make read mapping, variant calling, and genotyping variant-aware. Only recently a framework for variation graphs, vg [Garrison E, Adam MN, Siren J, et al. Variation graph toolkit improves read mapping by representing genetic variation in the reference. Nat Biotechnol 2018;36:875–9], have improved variation-aware alignment and variant calling in general. The major bottleneck of vg is its high cost of reads mapping to a variation graph. In this paper we study the problem of SNP calling on a variation graph and we present a fast reads alignment tool, named VG SNP-Aware. VG SNP-Aware is able align reads exactly to a variation graph and detect SNPs based on these aligned reads. The results show that VG SNP-Aware can efficiently map reads to a variation graph with a speedup of 40× with respect to vg and similar accuracy on SNPs detection.

Connecting transcriptional and post-transcriptional regulatory networks solves an important puzzle in the elucidation of gene regulatory mechanisms. To decipher the complexity of these connections, we build co-expression network modules for mRNA as well as miRNA expression profiles of breast cancer data. We construct gene and miRNA co-expression modules using the weighted gene co-expression network analysis (WGCNA) method and establish the significance of these modules (Genes/miRNAs) for cancer phenotype. This work also infers an interaction network between the genes of the turquoise module from mRNA expression data and hubs of the turquoise module from miRNA expression data. A pathway enrichment analysis using a miRsystem web tool for miRNA hubs and some of their targets, reveal their enrichment in several important pathways associated with the progression of cancer.

Biclustering is a non-supervised data mining technique used to analyze gene expression data, it consists to classify subgroups of genes that have similar behavior under subgroups of conditions. The classified genes can have independent behavior under other subgroups of conditions. Discovering such co-expressed genes, called biclusters , can be helpful to find specific biological features such as gene interactions under different circumstances. Compared to clustering, biclustering has two main characteristics: bi-dimensionality which means grouping both genes and conditions simultaneously and overlapping which means allowing genes to be in more than one bicluster at the same time. Biclustering algorithms, which continue to be developed at a constant pace, give as output a large number of overlapping biclusters. Visualizing groups of biclusters is still a non-trivial task due to their overlapping. In this paper, we present the most interesting techniques to visualize groups of biclusters and evaluate them.

Differential network analysis has become a widely used technique to investigate changes of interactions among different conditions. Although the relationship between observed interactions and biochemical mechanisms is hard to establish, differential network analysis can provide useful insights about dysregulated pathways and candidate biomarkers. The available methods to detect differential interactions are heterogeneous and often rely on assumptions that are unrealistic in many applications. To address these issues, we develop a novel method for differential network analysis, using the so-called disparity filter as network reduction technique. In addition, we propose a classification model based on the inferred network interactions. The main novelty of this work lies in its ability to preserve connections that are statistically significant with respect to a null model without favouring any resolution scale, as a hard threshold would do, and without Gaussian assumptions. The method was tested using a published metabolomic dataset on colorectal cancer (CRC). Detected hub metabolites were consistent with recent literature and the classifier was able to distinguish CRC from polyp and healthy subjects with great accuracy. In conclusion, the proposed method provides a new simple and effective framework for the identification of differential interaction patterns and improves the biological interpretation of metabolomics data.

Ovarian cancer is the third leading cause of cancer-related deaths in India. Epigenetics mechanisms seemingly plays an important role in ovarian cancer. This paper highlights the crucial epigenetic changes that occur in POTEE that get hypomethylated in ovarian cancer. We utilized the POTEE paralog mRNA sequence to identify major motifs and also performed its enrichment analysis. We identified 6 motifs of varying lengths, out of which only three motifs, including CTTCCAGCAGATGTGGATCA, GGAACTGCC, and CGCCACATGCAGGC were most likely to be present in the nucleotide sequence of POTEE. By enrichment and occurrences identification analyses, we rectified the best match motif as CTTCCAGCAGATGT. Since there is no experimentally verified structure of POTEE paralog, thus, we predicted the POTEE structure using an automated workflow for template-based modeling using the power of a deep neural network. Additionally, to validate our predicted model we used AlphaFold predicted POTEE structure and observed that the residual stretch starting from 237-958 had a very high confidence per residue. Furthermore, POTEE predicted model stability was evaluated using replica exchange molecular dynamic simulation for 50 ns. Our network-based epigenetic analysis discerns only 10 highly significant, direct, and physical associators of POTEE. Our finding aims to provide new insights about the POTEE paralog.

Glioblastoma is the most aggressive type of brain tumors resistant to a number of antitumor drugs. The problem of therapy and drug treatment course is complicated by extremely high heterogeneity in the benign cell populations, the random arrangement of tumor cells, and polymorphism of their nuclei. The pathogenesis of gliomas needs to be studied using modern cellular technologies, genome- and transcriptome-wide technologies of high-throughput sequencing, analysis of gene expression on microarrays, and methods of modern bioinformatics to find new therapy targets. Functional annotation of genes related to the disease could be retrieved based on genetic databases and cross-validated by integrating complementary experimental data. Gene network reconstruction for a set of genes (proteins) proved to be effective approach to study mechanisms underlying disease progression. We used online bioinformatics tools for annotation of gene list for glioma, reconstruction of gene network and comparative analysis of gene ontology categories. The available tools and the databases for glioblastoma gene analysis are discussed together with the recent progress in this field.